Skip to main content

Table 1 The features selected by the iterative genetic search algorithm with evaluation by Logistic Regression (LR), Baysian Network (BN), Functional Tree (FT), REP Tree (RT), and Alternating Decision Tree (AT)

From: Transient protein-protein interface prediction: datasets, features, algorithms, and the RAD-T predictor

Feature

Count

LR

BN

FT

RT

AT

relSESA

10

□∙

□∙

□∙

□∙

□∙

esolv

6

â–¡

□∙

∙

 

□∙

Density

6

∙

 

∙

□∙

□∙

ePot

5

 

□∙

â–¡

â–¡

∙

Scorecons

5

□∙

∙

â–¡

â–¡

 

rate4site

5

â–¡

∙

∙

â–¡

 

Disorder

5

□∙

 

∙

â–¡

â–¡

B-Factor

4

 

∙

□∙

 

â–¡

Roughness

4

□∙

□∙

   

Hydro

3

∙

 

â–¡

â–¡

 

Protrusion

3

□∙

  

â–¡

 

Propensity

3

â–¡

 

â–¡

∙

 

Curvature

2

 

□∙

   
  1. A white box (□) represents the feature was selected when the algorithm was tested on all proteins, using leave-one-out cross-validation, while a feature with a black circle (∙) was selected when tested on the NI1 subset. Count is the number of times a feature was selected by either datasets tested.